1  

An Integrated Study of Flue Gas Flow and Superheating Process in Recovery Boiler using Computational Fluid Dynamics and 1D-Process Modelling

Kunal Kumar Andritz Oy, Viljami Maakala Andritz Oy and Ville Vuorinen Aalto University

ABSTRACT

Superheaters are the last heat exchangers on the steam side in recovery boilers. They are typically made of expensive materials due to the high steam temperature and risks associated with ash-induced corrosion. Therefore, detailed knowledge about the steam properties and material temperature distribution is essential for improving the energy efficiency, cost efficiency and safety of recovery boilers. In this work, for the first time, a comprehensive 1D-process model (1D-PM) for superheated steam cycle is developed and linked with a full-scale 3D CFD (computational fluid dynamics) model of the superheater region flue gas flow. The results indicate that first; the geometries of headers and superheater platens affect platen-wise steam distribution. Second, the CFD solution of the 3D flue gas flow field and platen heat flux distribution coupled with the 1D-PM affect the generated superheated steam properties and material temperature distribution. These new observations clearly demonstrate the value of the present integrated CFD/1D-PM modelling approach. It is advantageous for trouble shooting, optimizing the performance of superheaters and selecting their design margins for the future. The developed integrated modelling approach could also be relevant for other large-scale energy production units such as biomass-fired boilers.

1 INTRODUCTION

Recovery boilers are used to combust black liquor for chemical recovery and producing high-pressure superheated steam. The generated steam is utilized for self-sustainable pulp mill operations and electricity generation. For instance, in Finland in 2017, 8.1% of total electricity was generated with black liquor combustion in recovery boilers [1]. Global production of chemical wood pulp has been forecasted to increase annually by 1% [2]. Tran et al. [3] noted that approximately 1.5 kg of black liquor dry solids (BLDS) are produced per 1.0 kg of chemical wood pulp production and around 3.5 kg of superheated steam is generated per 1.0 kg of BLDS combustion in recovery boilers. Hence, black liquor is a vital biomass based renewable energy source from future perspective.

Over recent years, the global interest for carbon neutral energy production has been continuously increasing to mitigate the climate change. Simultaneously, the conventional role of recovery boilers as chemical recovery units is shifting towards renewable energy production [4]. In addition, the average capacity of recovery boilers has been increasing. The current largest capacity of a recovery boiler is 12000 tds/d and even larger recovery boilers have been planned. Therefore, it is essential to develop new computational models for such large boilers to understand their heat transfer phenomena in detail, and to improve their contribution for renewable energy production.

The superheater region in a recovery boiler is the focus of this work. The superheaters are single-phase heat exchangers. They are used to convert saturated steam into superheated steam by capturing heat from hot flue gas (≈ 30% of total). They are the last and one of the largest heat transfer surfaces in a recovery boiler before the steam turbine. Thence, the optimal performance of superheaters, including higher quality superheated steam production and reduction in material issues such as corrosion, is essential for efficient and safe recovery boiler power plant operation.

Previously, detailed studies have been performed for coal-fired boilers and bubbling fluidized bed (BFB) boilers to analyze the heat transfer sections including superheaters, and to study the steam generation process using integrated CFD/1D-PM modelling approaches. In integrated CFD/1D-PM simulations, a flue gas side 3D CFD (computational fluid dynamics) model is coupled with a 1D-PM (1D-process model) of water-steam side. It is beneficial to utilize the 1D-PM for large and complex flows in the steam cycle, which are not feasible to solve with standalone CFD modelling. The reasons are large computational cost, time and availability of computational resources.

2  

Edge et al. [5] studied the steam generation process in a 500 MWe natural circulating coal-fired boiler using integrated CFD/1D-PM simulations. Schuhbauer et al. [6], Chen et al. [7] and Park et al. [8] performed integrated CFD/1D-PM simulations, and studied the heat transfer between hot flue gas and water-steam cycle in coal-fired boilers. Yang et al. [9] also carried out integrated simulations to analyze the temperature distribution on furnace walls and the heating process of supercritical- carbon dioxide (S-CO ) in a conceptual higher efficiency (≥50%) coal-fired boiler. Moreover, Hovi et al. [10] carried out transient integrated simulations, and investigated the effects of rapid load change situations on flue gas temperature, heat transfer and pollutant formation in a BFB boiler. Hence, it is seen that integrated modelling approach is state-of-the-art for coal-fired boilers and BFB boilers. However, in these previous studies the 1D-process models have mainly focused on the water circulation process, and the models utilized for the heat transfer sections have been simplified and based on porous media method. The porous media method does not provide an accurate solution for the flow field and material temperature distribution for the tube bundles in the heat transfer section. Therefore, in the present work, each superheater platen is modeled separately and comprehensively on both the CFD side and the 1D-PM side, which has not been done previously to the authors’ knowledge.

In context of recovery boilers, the flue gas flow field and heat transfer in superheater region have been studied previously using standalone CFD simulations. Saviharju et al. [11] analyzed the flow field and temperature distribution in the upper furnace for two recovery boilers. Leppänen et al. [12, 13, 14, 15] studied deposit formation in recovery boilers and compared the results with experimental data. Maakala et al. [16] used surrogate-based analysis with CFD to optimize the heat transfer in superheater region. Maakala et al. [17] developed a detailed 3D CFD model for superheater region, and obtained a detailed 3D solution for flue gas flow field and heat flux distribution to superheater platens. However, the effects of flue gas side on the steam cycle and vice versa have not been well explored in the superheater region of recovery boilers, even though recovery boilers contribute around 25% of global industrial biomass based energy production [18]. In addition, there are few previous studies available where full-scale 3D CFD modelling approach has been adopted for recovery boiler simulations.

Therefore, the main objective of this paper is to improve the understanding about heat transfer between hot flue gas and superheated steam cycle. The study includes the effects of 3D flue gas flow field in superheater region on heat flux distribution, steam distribution and material temperature distribution among the superheater platens. For this purpose, a full-scale 3D CFD model of superheater region is coupled with a detailed 1D-PM and integrated CFD/1D-PM simulations are performed. The developed 1D-PM is validated with reference data. This integrated CFD/1D-PM modelling approach is the novelty of this work.

2 METHODS AND MODELS

2.1 Description of the Case

Figure 1-a shows the domain of the recovery boiler CFD model. The superheater region is marked in the figure by a rectangular box. The capacity of the recovery boiler is 1000 tds/d. The combustion of black liquor is assumed to be completed before the flue gas reaches the superheater region. Therefore, the furnace is not considered in this work. Table 1 shows the main operating parameters of the boiler. The reference data for recovery boiler was obtained at approximately 80% of its total capacity. It comprises of mass and energy balance calculations, and data from a measurement campaign. The model inlet is located between the tertiary air supply level and nose arch. This is done to assure that the tertiary air supply has minimum effect on the flue gas flow and the flow field is steady when flue gas reaches the superheater region. Similarly, the outlet is located far away from the superheater region to prevent the impact of outlet boundary conditions to the numerical solution of superheater region.

The boiler walls, rear wall screen and boiler bank are evaporating surfaces. They are used to convert saturated water into saturated steam at almost a constant saturation temperature. The chosen recovery boiler has four stages of superheating (SH) including SH1A, SH1B, SH2, SH3 and SH4. The first stage superheaters (SH1A and SH1B) are counter-current superheaters according to the flue gas flow direction. The other superheaters are co-current heat

3  

exchangers. Each superheater is made of 21 platens that are equally spaced across the width of the boiler. In reality, each platen has in-line, thin, seamless and tightly spaced tubes that carry steam inside. In this work, the superheater platens are considered as flat plates and linked with the 1D-PM on the steam side. Similarly, the boiler walls are modelled as flat surfaces instead of tightly fitted heated riser tubes. The boiler bank is modelled as a porous medium with calculated porosity, inertial loss coefficients and heat sink values. These simplifications are used to reduce the calculation time and complexity of the integrated CFD/1D-PM simulations.

a

b

Figure 1: a) A two dimensional view of the recovery boiler geometry. 1) furnace (not considered), 2) inlet, 3-7) boiler walls, 8) nose level, 9) rear wall screen, 10) boiler bank, 11) outlet, 12) steam inlet and 13) superheated steam to the steam turbine. The boiler walls 3 and 4 are the side walls. The superheater region is represented with a rectangular box. b) The meshed geometry of the recovery boiler.

Table 1: Main operating values for the recovery boiler. All the black liquor values are virgin dry solids values.

Parameters Values

Boiler type Kraft recovery boiler

Black liquor capacity, tds/d 1000

Black liquor higher heating value, MJ/(kgds) 15

BLDS, % 74

Main steam flow (m), kg/s 49

Main steam temperature (T), °C 505

Main steam pressure (P), bar 110

4  

2.2 CFD Modelling  

Figure 1-b shows the computational grid for the present domain. It is discretized using a polyhedral meshing approach and it consists of approximately 13M polyhedral cells. The generated mesh has a very fine resolution for superheater platens. This is done to assure that the calculation nodes (discretized elements) of each superheater tube in the 1D-PM can precisely couple with a certain number of faces on the walls of the corresponding superheater platens in the CFD model. The proper coupling of 1D-PM-side calculation nodes and CFD-side faces is essential for accurate integrated CFD/1D-PM simulations. Additionally, an adequate number of cells are placed in the grid between superheater platens to accurately solve the flue gas flow field and heat transfer phenomena.

The present CFD model solves the fundamental equations of fluid dynamics, turbulence, species transport and radiation in steady state, Reynolds-averaged form and incompressible flow conditions using ANSYS Fluent 18.1. The pressure-based solver is used and segregated SIMPLE scheme is applied for pressure velocity coupling. The standard

model with standard wall functions is utilized for turbulence modelling. The species transport equations are solved for flue gas species including H O (gaseous water), CO (carbon dioxide), O (oxygen) and N (nitrogen). The flue gas species N and O are diathermanous in nature and do not contribute in radiation. Whereas, the species CO and H O emit and absorb radiation at small wavelength bands. Therefore, the non-gray weighted sum of gray gases method with five wavelength bands is utilized with the Discrete Ordinates radiation model. A model based on Wessel et al. [19] is used to solve the effect of fume particles (aerosol particles) on radiative properties of flue gas. Deposition on boiler walls, superheater platens and rear wall screen is considered with fixed deposit values.

The boundary conditions at domain inlet are taken from a previously performed CFD simulation of black liquor combustion in the furnace. These inlet boundary conditions are flue gas velocity, temperature, turbulence properties and species mass fractions. The thermal boundary conditions on the walls except superheater platens are given as convective heat transfer boundary conditions by setting the overall heat transfer coefficient ( ) and free-stream temperature ( ). The thermal boundary conditions for superheater platens are described in Section 2.4.

The total heat flux ( ) on a wall is

(1)

1/1

(2)

where is convective heat transfer coefficient on flue gas side, is wall (or deposit) surface temperature, is flue gas temperature, is radiative heat flux, is deposit thickness, is deposit thermal conductivity,

is superheater tube thickness, is superheater tube thermal conductivity and is water-side heat transfer coefficient.

In reality, the deposition properties are hard to estimate in recovery boilers. According to literature, deposit thickness and its thermal conductivity in recovery boilers are in the range of 5-60 mm and 0.1-2.5 W/(mK) respectively (Leppänen et al. [13], Maakala et al. [17], Li et al. [20] and Zbogar et al. [21]). Due to the uncertainty involved, the overall heat transfer coefficients ( ) are fitted to reference data, similarly as has been done by Leppänen et al. [13] and Maakala et al. [16, 17]. The value of is chosen as 1 W/(mK). Table 2 shows the deposit thickness, inlet boundary conditions and thermal wall boundary conditions.

5  

Table 2: The inlet and wall boundary conditions for the CFD model. The inlet boundary conditions are presented in terms of average values. For superheaters, and are calculated during integrated CFD/1D-PM simulations.

Parameters Values Inlet boundary conditions

Velocity, m/s 4.65 Temperature, °C 932 Flue gas mass flow rate, kg/s 56.78 Reynolds number 175000 Flue gas composition (wt %) Carbon dioxide (CO Gaseous water (H O Oxygen (O Nitrogen (N

21 15

2 62

Wall thermal boundary conditions Walls W

m K

K mm

Boiler walls and boiler bank walls 28.3 599 35 Rear wall screen 610 599 1.2 SH1A - - 1.0 SH1B - - 3.5 SH2 - - 13.5 SH3 - - 8.0 SH4 - - 6.7

2.3 1D-PM Modelling

Figure 2 shows the steam cycle for superheater region. It comprises of steam drum, inlet headers, outlet headers and superheater platens. The water-steam mixture from the evaporating surfaces is collected into steam drum where the saturated steam is separated from the mixture. The saturated steam is then sent to superheaters for increasing its temperature to the required outlet temperature. The steam side 1D-PM for the superheater region is developed using Apros 6. The headers, connecting pipes and steam flow loops of superheater platens are modelled in full detail.

The thermal-hydraulics properties of single-phase steam flow in superheater tubes are solved using a homogenous (three-equation) model. This model solves the conservation equations for mass, momentum and energy for superheated steam in Z-direction. The pressure losses in superheater tubes are mainly caused by pipe friction and minor losses or form losses due to the geometrical structure of the piping system [22, 23]. The total pressure loss (Δ ) in a pipe flow is calculated as

Δ2

(3)

where is the friction factor, is the pipe length, is inner diameter of pipe and ∑ is the sum of all form loss coefficients in the piping system. The flow boundary conditions for the 1D-PM are shown in Table 3.

6  

Figure 2: The superheater region steam cycle. 1) steam drum, 2-15) inlet and outlet headers. The headers are connected in cross-patterns using connecting pipes. The final superheated steam is sent to the steam turbine using main steam pipe. The main steam pipe is connected to SH4 outlet header (14-15).

Table 3: The boundary conditions for the 1D-PM and properties of injected water in between superheating stages. The boundary conditions are based on reference data.

Parameters P (bar) T (°C m (kg/s) Inlet / steam drum 121.9 325.9 - Outlet / main steam pipe Adjusted 506 38.2 Desuperheating stages Pressure and temperature for each stage 124.9 140.5 - SH1-SH2 - - 0.18 SH2-SH3 - - 0.62 SH3-SH4 - - 0.26

2.4 Integrated CFD/1D-PM Modelling

The flue gas side 3D CFD model is coupled with the steam side 1D-PM using a two-way heat transfer coupling method. This method is applied to superheater platens. In this approach, the CFD side faces of an individual platen are mapped with particular calculation nodes of superheater tubes in the 1D-PM. It is achieved by linking the coordinate systems of both calculation models. As an example, the coupled CFD faces and 1D-PM calculation nodes for a superheater platen are shown in Figure 3-a.

During the integrated CFD/1D-PM simulations, the 1D-PM calculates the deposit temperature ( ) at the surfaces of superheater platens and sends it to the CFD model. The CFD model then determines the surface heat transfer rate ( ) and transfers it to the coupled calculation nodes of the 1D-PM. These thermal wall boundary conditions for superheater platens are exchanged at every CFD iteration. An example is shown in Figure 3-b.

7  

a

b

Figure 3: a) Coupled CFD faces and 1D-PM calculation nodes for a superheater platen. The legend describes the indices of mapped calculation nodes. Therefore, there are 388 nodes in this superheater platen. Indirectly, the figure also shows the superheater tubes. b) The exchange parameters during integrated CFD/1D-PM simulations. One superheater tube is presented. N1-N3 are the calculation nodes of heat pipe or superheater tube. N4-N6, N7-N9 and N10-N12 are the heat structure nodes for steam temperature, tube material temperature and deposit layer temperature calculations. F1-F3 are the CFD faces, which are coupled with calculation nodes (N10-N12) of 1D-PM. The exchange values are temperature (T), from 1D-PM to CFD, and surface heat transfer (q), from CFD to 1D-PM.

8  

3 RESULTS AND DISCUSSION

3.1 Validation of 1D-PM Modelling Approach

The consistency and accuracy of the developed 1D-PM are analyzed by comparing its results with reference data. This is done before performing the integrated CFD/1D-PM simulations. The boundary conditions for the 1D-PM are shown in Tables 3 and 4 that are based on reference data. The thermal wall boundary conditions to superheater platens are given as a uniform heat flux distribution, which is a common approach when no more detailed information is available.

The computed results of 1D-PM are in good agreement with reference data as is shown in Table 5. The calculated pressure losses (Δ ) and steam superheating (Δ ) across the superheaters deviate from reference data by maximum of 9% and 3% respectively. The main steam mass flow rate calculated by 1D-PM is similar to reference data. However, the main steam pressure and temperature deviate by 1.4% and 1% respectively. The main reasons for above mentioned discrepancies are pipe friction and form losses due to the complex geometry of connecting pipes, headers, steam flow loops in superheater platens and main steam pipe. Therefore, based on this validation study, the developed 1D-PM is considered to be consistence with good accuracy.

Table 4: Heat flux distribution to superheater platens based on reference data.

Superheaters (kW) Platens (kW) (kW/m ) SH1A 3814 21 181.62 4.75 SH1B 3566 21 169.81 5.71 SH2 9773 21 465.38 9.89 SH3 8209 21 390.90 7.72 SH4 3017 21 143.67 3.75

Table 5: Comparison between reference data and developed 1D-PM for validation study.

Pressure losses and steam superheating for superheatersSuperheaters Δ (bar) Δ (°C) Δ (bar) Δ (°C) SH1A 0.41 14 0.43 13.7 SH1B 0.35 18 0.38 18.1 SH2 2.20 69 2.28 68.1 SH3 2.04 73 2.14 71.8 SH4 2.60 28 2.77 27.2

Main steam properties Parameters Reference Data 1D-PM Relative error (%) P, bar 111.9 110.29 -1.4 T, °C 506 501.3 -1.0 m, kg/s 38.2 38.2 -

3.2 Integrated CFD/1D-PM Simulations

3.2.1 Flue gas side.

Figure 4-a shows the flue gas flow field in the middle of superheater region. Three recirculation zones (1, 2 and 3) are identified. This kind of vortex structures at different locations are also noted in other recovery boiler simulations such

9  

as Saviharju et al. [9] and Maakala et al. [15, 16]. The smaller recirculation zones 2 and 3 are located in the corner of front cavity and below the SH4 platens respectively. The larger recirculation zone (LRZ) (1) is located in the middle of superheater region. It extends mainly from SH2 platens to SH4 platens across the boiler width and depth. The observations indicate that the partial boiler load (80%) and uneven inlet velocity profile are responsible for the occurrence of these vortex structures. Engblom et al. [24] also noted the effect of partial furnace load on asymmetries in flow field in recovery boiler using both measurements and CFD simulations.

Figure 4-b shows the flue gas temperature field in the middle of superheater region. The vortex structures, especially LRZ, significantly affect the flue gas temperature field. The flue gas temperature in LRZ is in the range of 440-530°C, which is lower than the surrounding flue gas temperature. The surface areas of superheater platens in this zone, therefore, are inefficiently used for heat transfer. Hence, the uneven flow field in superheater region is connected to variations in platen-wise generated steam properties and material temperature distribution, which are analyzed in Section 3.2.2.

a

b

Figure 4: a) The solved velocity field for flue gas. b) The solved flue gas temperature field. 1, 2 and 3 represent the recirculation zones. Figures are taken from the middle of the boiler width.

The heat flux distribution to superheater platens in integrated CFD/1D-PM simulations is shown in Figure 5. In the figure, the uniform platen-wise heat flux distribution for standalone 1D-PM simulation is also shown for reference because it is a common assumption when no more detailed information is available.

The 3D flow field in superheater region substantially affects the platen-wise heat flux distribution for superheaters. The LRZ in the middle of superheater region leads to lower the heat flux on the middle platens compared to platens near side walls as shown in Figure 5. The largest differences for platen-wise heat flux distribution are noted for SH4 and SH1A, where the heat fluxes on the platens near side walls are respectively 83% and 80% higher than the platens in the middle region.

10  

a

b

c

d

e

Figure 5: Heat flux distribution to superheater platens in integrated CFD/1D-PM simulations. The platen-wise uniform heat flux distribution for standalone 1D-PM simulation is shown for reference. The right wall and left wall represent the side walls of recovery boiler.

3.2.2 Steam side.

The steam side results for integrated CFD/1D-PM simulations and their comparison with standalone 1D-PM simulation are discussed in this section. The comparison study is performed to explicitly show the effect and advantages of integrated modelling approach over standalone 1D-PM simulation. For the purpose of this comparison, the total heat transfer to each superheater in standalone 1D-PM simulation was set to be the same as in integrated CFD/1D-PM simulations. In this paper, for brevity, the comparison results for SH1A and SH4 are mainly discussed.

11  

The main steam values including pressure, temperature and mass flow rates in integrated simulations and standalone 1D-PM simulation are close to each other with negligible deviations. For both simulations, these values are approximately 110.2 bar, 504 °C and 38.2 kg/s. Figure 6 shows the total outlet steam mass flow rate from each superheater. For both integrated CFD/1D-PM and standalone 1D-PM simulations the geometrical structure of main steam pipe causes variation in outlet steam mass flow rates for SH4 outlet header.

The platen-wise pressure losses, steam distribution and steam temperature for SH1A and SH4 are shown in Figure 7. The pressure losses calculated in integrated CFD/1D-PM and standalone 1D-PM simulations are close to each other with small discrepancies. For both simulations, the maximum differences of 2.4% and 0.8% in platen-wise pressure losses are found for SH1A and SH4 respectively. The deviations in pressure losses for SH1B, SH2 and SH3 are 6%, 1.86% and 1.5% respectively.

Moreover, the 3D heat flux distribution in superheater region has a smaller impact on platen-wise steam distribution compared to the pressure losses caused by the complex geometry of superheated steam cycle. The comparison study shows that maximum differences between platen-wise steam distribution for SH1A and SH4 are 2.3% and 0.56%. Whereas, they are 2.3%, 2.2% and 1.09% for SH1B, SH2 and SH3 respectively. For integrated CFD/1D-PM simulations, the deviations between minimum and maximum platen-wise steam mass flow rates are in the range of 3%-7%.

However, the non-uniform 3D heat flux distribution in superheater region has substantial effects on platen-wise generated steam temperatures. For SH1A and SH4, the platens near side walls have higher steam temperatures compared to platens in middle region as these platens receive higher heat fluxes (see Figure 7). Similar behavior is also observed for other superheaters. For integrated CFD/1D-PM simulations, the deviations in platen-wise superheating are in the range of 45%-122%. On the contrary, the standalone 1D-PM simulation provides almost uniform platen-wise steam temperatures for the superheaters. Therefore, it is considered that the superheated steam generation process based on uniform heat flux distribution approach is not an accurate method, as it does not consider the effects of flow field in superheater region.

The platen-wise material temperature distribution for outer (shortest) and innermost (longest) steam flow loops in SH1A and SH4 are shown in Figure 8. Similarly to platen-wise steam temperature distribution, the standalone 1D-PM simulation provides almost uniform and most likely inaccurate results. It is considered that in reality the non-uniform platen-wise heat flux distribution in superheater region should be accountable for variation in flow loop-wise material temperature distribution. In fact, the integrated CFD/1D-PM simulations are able to capture these complex phenomena as is shown in Figure 8. Moreover, the results of integrated CFD/1D-PM simulations are also compared with measurement data, which show similar trends in material temperatures. The deviations are in the range of 1-6% (Table 6). For all the superheaters, the average differences between measurements and results of integrated CFD/1D-PM simulations are between 0.7%-2.6% (Table 6). The variation in flow field in superheater region during measurements and integrated CFD/1D-PM simulations is considered to be mainly responsible for these discrepancies.

Figure 6: The outlet steam mass flow rates from the exits including left wall side and right wall side of outlet headers for all the superheaters.

12  

a

b

c

d

e

f

Figure 7: Platen-wise pressure losses, mass flow rates and steam temperature for SH1A (a, b and c) and SH4 (d, e and f).

13  

a

b

c

d

Figure 8: Platen-wise material temperature distribution of SH1A (a and b) and SH4 (c and d). The outermost (shortest) flow loop is represented as L1 whereas and L3 (SH4) and L4 (SH1A) represent the innermost (longest) flow loops.

Table 6: Maximum and average differences between measurement data and results of integrated CFD/1D-PM simulations for superheater material temperature distribution.

Maximum difference Superheaters Platen number Flow loop number Maximum error (%) SH1A 7 3 1.6 SH1B 7 1 2.5 SH2 5 3 3.2 SH3 10 3 5.7 SH4 10 3 5.0

Average difference Superheaters Average difference (%) SH1A 0.7 SH1B 1.0 SH2 1.5 SH3 2.2 SH4 2.6

14  

4 CONCLUSIONS

The developed integrated CFD/1D-PM modelling approach was demonstrated to be feasible for solving the complex heat transfer phenomena between steam and flue gas in the superheater region with good accuracy. In comparison to previous approaches (porous media method [5-10] and 3D slice superheater region method [11-17]), the relevant flow and heat transfer phenomena are captured on a much more detailed level. The integrated modelling approach helps to obtain significantly more accurate solutions for the platen-wise superheated steam distribution, steam temperature and material temperature distribution. The improved accuracy is enabled by the detailed modelling of the steam side geometry and process as well as the coupled solution of the flue gas flow and heat transfer.

The integrated CFD/1D-PM modelling approach provides a novel way to understand the superheating process in detail and to optimize the future design of superheaters. It also helps to improve the safety and thermal efficiency of recovery boilers. It could also be applied to other energy production units such as biomass-fired boilers and utility boilers.

Based on the results, the following future research directions are identified:

1. Full-scale time-dependent integrated CFD/1D-PM simulations including black liquor combustion in lower furnace will be performed to further investigate the superheated steam generation process in a more precise way. This full-scale integrated modelling also corresponds more accurately to real recovery boiler operation. With this approach, rapid load change situations can also be studied.

2. For inlet and outlet headers, a CFD study will be performed to precisely understand the effects of their geometries on steam distribution. It will help to explore new possibilities for optimizing their performance and design.

5 REFERENCES

[1] Official Statistics of Finland (OSF): Production of electricity and heat [e-publication]. ISSN=1798-5099. Helsinki: Statistics Finland [referred: 11.12.2018]. Access method: http://www.stat.fi/til/salatuo/index_en.html. [2] Pöyry, World Fibre Outlook up to 2030, 2016. [Electronic publication]. [3] Tran, H., and Vakkilainen, E., 3rd ICEP international colloquium on eucalyptus pulp, Belo Horizonte, Brazil, 2007, p. 4. [4] Salmenoja K., 55th Anniversary International Recovery Boiler Conference, Turku, Finland, 2019, p. 175. [5] Edge, P. J., Heggs, P. J., Pourkashanian, M., and Williams, A., Computers & Chemical Engineering 35(12): 2618(2011).

[6] Schuhbauer, C., Angerer, M., Spliethoff, H., Kluger, F., and Tschaffon, H., Fuel 122: 149(2014).

[7] Chen, T., Zhang, Y. J., Liao, M. R., and Wang, W. Z., Fuel 240:49(2019).

[8] Park, H. Y., Faulkner, M., Turrell, M. D., Stopford, P. J., and Kang, D. S., Fuel 89(8): 2001(2010).

[9] Yang, Y., Bai, W., Wang, Y., Zhang, Y., Li, H., Yao, M., and Wang, H., Applied Thermal Engineering 113: 259(2017).

[10] Hovi, V., Huttunen, M., Karppinen, I., Pättikangas, T., Niemistö, H., Karvonen, L., Kallio, S., Tuuri, S., and Ylä-Outinen, V. Energy Procedia 120: 508 (2017).

15  

[11] Saviharju, K., Pakarinen, L., Wag, K., and Välipakka, I., International Chemical Recovery Conference, TAPPI PRESS, Atlanta, GA, USA, 2004, p. 247.

[12] Leppänen, A., Välimäki, E., Oksanen, A., and Tran, H., TAPPI Journal 12(3): 25(2013).

[13] Leppänen, A., Tran, H., Taipale, R., Välimäki, E., and Oksanen, A., Fuel 129: 45(2014).

[14] Leppänen, A., Tran, H., Välimäki, E., and Oksanen, A., Journal of Science & Technology for Forest Products and Processes 4(1): 50(2014).

[15] Leppänen, A., and Välimäki, E., TAPPI Journal 15(3): 187(2016).

[16] Maakala, V., Järvinen, M., and Vuorinen, V., Energy 160: 361(2018).

[17] Maakala, V., Järvinen, M., and Vuorinen, V., Applied Thermal Engineering 139: 222(2018).

[18] Vakkilainen, E., Kuparinen, K., and Heinimö, J., IEA Bioenergy Task 40. 2013. 2013p.

[19] Wessel, R. A., Denison, M. K., and Samretvanich, A., TAPPI Journal 83(7): 1(2000).

[20] Li, B., Brink, A., and Hupa, M., Fuel Processing Technology 105:149(2013).

[21] Zbogar, A., Frandsen, F. J., Jensen, P. A., and Glarborg, P., Progress in Energy and Combustion Science 31(5-6): 371(2005).

[22] Alobaid, F., Mertens, N., Starkloff, R., Lanz, T., Heinze, C., and Epple, B., Progress in Energy and Combustion Science 59: 79(2017).

[23] Faculty of Engineering and Applied Science, Mechanical and Materials Engineering, Queen’s University, Canada. Available at: https://me.queensu.ca/People/Sellens/LossesinPipes.html.

[24] Engblom, M., Miikkulainen, P., Brink, A., and Hupa, M., TAPPI Journal 11(11): 19(2012).

Gateway to the Future

An Integrated Study of Flue Gas Flow and Superheating Process in Recovery Boiler using Computational Fluid Dynamics and 1D-Process

Modelling

Kunal Kumar

ANDRITZ OY, Finland

CHAPTER OVERVIEW

01 INTRODUCTION

02 MOTIVATION

03 MODELLING APPROACH

04 INTEGRATED CFD/1D-PM SIMULATIONS

05 RESULTS AND DISCUSSION

06 CONCLUSIONS

INTRODUCTION

Superheaterplaten

Superheaterregion

Superheater region is the focus of this research work.

MOTIVATION

Increase recovery boiler´s

availability to support

continuous pulp production.

Obtain detailed knowledge

about :-a)Platen-wise superheated steam

distribution.

b)Platen-wise superheater material temperature distribution.

Improve future design

(Food and AgriculturalOrganization of UnitedNations (FAO))

MODELLING APPROACHDescription

Parameters Values

Boiler type Kraft recoveryboiler

Black liquor capacity, tds/d

1000

Black liquor HHV, MJ/(kgds)

15

BLDS, % 74

Main steam flow, kg/s 49

Main steam temperature, °C

505

Main steam pressure, bar

110

Main operating parameters

Reference Data: Mass and energybalance calculations, andmeasurement. It is based on 80%of boiler capacity.

MODELLING APPROACHCFD side: Flue gas

CAD model Computational domain (13M polyhedral cells

MODELLING APPROACHSteam side: 1D-PM

MODELLING APPROACHSteam side: 1D-PM

SH1A&1B platens

Inlet header

Outlet headerSteam

De-superheatingwater

Steam inlet Steam outle

Steam flow loops

T-Junction

Inlet header

Outlet header

Cross flow at SH1A inlet

INTEGRATED CFD/1D-PM SIMULATIONS

Temperature

Heat flow

CFD side meshed geometry Steam side (1D-PM)

Heat pipe calculation viewSuperheater tube

Platen wallCoupled 1D-PM

nodes and CFD faces for a platen

Boundaryconditionsexchange ateach CFDiteration 3

INTEGRATED CFD/1D-PM SIMULATIONSMethods and boundary conditions

Integrated model

solves:

a)Turbulence and

Radiation.

b)Flue gas species

transport.

c)Heat transfer.

d)Thermal-hydraulics of

Flow boundary

conditions:

a)CFD side (Flue gas):

Taken from previously

performed furnace CFD

simulation.

b)Steam side (1D-PM):

Based on Reference Data.

RESULTS AND DISCUSSION1D-PM validation with Reference Data

0.41

0.35

2.2

2.04

2.6

0.43

0.38

2.28

2.14

2.77

0 0.5 1 1.5 2 2.5 3

SH1A

SH1B

SH2

SH3

SH4

Pressure loss (bar)

Superheaters

Pressure losses

1D‐PM

Reference data14

18

69

73

28

13.7

18.1

68.1

71.8

27.2

0 10 20 30 40 50 60 70 80

SH1A

SH1B

SH2

SH3

SH4

Superheating (°C)

Superheaters

Superheating

1D‐PM

Reference data

Reference Data

1D-PM Error (%)

P, bar

111.9 110.29

1.4

T, °C 506 501.3 1.0

m, kg/s

38.2 38.2 0

Uniform platen-wise heat flux distribution is applied, which is based on Reference Data.

Final steam values

RESULTS AND DISCUSSIONIntegrated simulations: Flue gas side

Heat transfer to superheaters.Flue gas flow and temperature fields in the middle of the boiler.

1- Large recirculation zone2&3- Small vortices

Non-uniform flue gas flow field, because of 1.Partial boiler load (80%).2 N if i l t b d

RESULTS AND DISCUSSIONIntegrated simulations: Steam side for SH3

Heat flux Mass flow rate

Steam temperature Steam superheating

Heat Flux for Standalone approach is taken from integrated simulations and uniformly distributed among the platens. It is a commonly used approach.

Final steam values

Integrated

Standalone

P, bar

110.17 110.24

T, °C 504 504.1

38 2 38 2

RESULTS AND DISCUSSIONIntegrated simulations: Steam side for SH3

L1≈L2< L3 ≈ L4

Superheaters Avg. diff. (%)

SH1A 0.7SH1B 1.0SH2 1.5SH3 2.2SH4 2.6

Measurements vs Integrated

Total 93 thermocouple measurements in all the superheaters.

CONCLUSIONS This novel Integrated CFD/1D-PM modelling approach is complex, comprehensive and advantageous. Its feasibility depends on preprocessing/ set-up, computational resources and available computational time.

Advantageso Able to solve complex heat transfer phenomena in superheater region of a recovery boiler.

o Aid for trouble shooting and optimizing the future design of superheaters.

o Relevant to study rapid load change situations in recovery boilers

Thank You!

Q & A